Dear FLUKA-experts,
Questions:
1) Does flair offer an option to limit the jobs that run at the same
time e.g. batches of 20 runs until they are completed?
2) Can someone share a simple batch script that allows me to use the
flair generated FLUKA .inp files and run them in batches with N runs at
the time until completion (ideally checking for the programm status)?
Is there a difference for running on a personal worksatation or cluster?
3) Does flair offer to limit the number of CPUs and / or RAM allocated
to FLUKA runs?
Background:
I am looping over a variable in flair. Thereby about 250 runs are
generated. Unfortunately, many timeout prior to completion i.e. the
status in flair is shown as "TIME-OUT" and the process is not running
when I checked htop. I run a personal workstation (8 core CPU; 16 GB
RAM; SL7 - 64-bit), no cluster etc.
From the documentation I understand that the START card offers WHAT(3)
and WHAT(6) to define time(outs?) for FLUKA runs.
http://www.fluka.org/fluka.php?id=man_onl&sub=73
If I understand correctly, WHAT(6) is the time reserved for an
interactive run that is only applicable in a few systems such as
clusters (?).
Therefore, I varied WHAT(3) but the runs still time-out when all are
started within flair at the same time. Individual runs or up to 20 runs
work out with no problem.
Thank you!
Best wishes,
Timo
__________________________________________________________________________
You can manage unsubscription from this mailing list at
https://www.fluka.org/fluka.php?id=acc_info
Received on Tue Sep 24 2019 - 15:35:03 CEST